A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 10, 16:30 UTC
[Cleanup] Remove redundant `decoder_layer_type` assignment in `Qwen2` (#31760)
Signed-off-by: maang <maang_h@163.com>
maang committed
Jan 6, 2026 at 10:09 UTC
cd1245a1848d866fa640cb1051c6bbb5b1c68f4a
1 file changed
-2
vllm/model_executor/models/qwen2.py
-2
@@ -389,8 +389,6 @@ class Qwen2Model(nn.Module):
389
else:
390
self.embed_tokens = PPMissingLayer()
391
392
- # Use the provided decoder layer type or default to Qwen2DecoderLayer
393
- decoder_layer_type = decoder_layer_type or Qwen2DecoderLayer
392
self.start_layer, self.end_layer, self.layers = make_layers(
393
config.num_hidden_layers,
394
lambda prefix: decoder_layer_type(